Spatial early warning signals discerned from prey fish behavior in whole-ecosystem regime shift experiment

Background/Question/Methods

Ecological regime shifts are rapid changes to a new state often occurring by surprise that may result in the loss of critical ecosystem services. In spatially coupled systems, regime shifts may be announced in advance by increasing spatial variance, increasing spatial autocorrelation, shifting spatial skewness, and shifts to low frequency spatial variance. Spatial early warning indicators are potentially important for application to management because information gained by sampling many points in space at one time may alleviate a need to collect long time series. While statistical indicators have proven effective in simulation models, there are few tests of spatial early warning indicators from real systems.

We evaluated spatial autocorrelation, spatial variance, spatial skewness, and the discrete Fourier transform of spatial data as early warning indicators by applying these metrics to spatially distributed prey fish catch data from a whole-ecosystem experiment designed to create a regime shift. We added top predators to a lake ecosystem to induce a trophic cascade. This manipulation was meant to directly reduce prey fish biomass and force them from open water habitats to near shore refuge. These predation avoidance behaviors should amplify the observed variability that is expected prior to the shift thus triggering spatial early warning indicators.

Results/Conclusions

Prey fish abundance declined by more than 80% through the transition. Early warning signals were present more than a year in advance of the transition and these signals were due to both changing fish biomass and behavior. Spatial variance, spatial skewness, and spatial autocorrelation all exhibited early warnings, but shifts to low frequency spatial variance were more powerful indicators of the regime shift in this experiment. These results offer a real world test of spatial early warning indicators, but are particularly important for advancing our ability to detect critical transitions in mobile animal populations such as fisheries.